# Filter Checkboxes vs a Guided Question Flow A/B Test

Page: https://alidemirbas.com.tr/en/lab/ab-test-library/tests/filter-checkboxes-vs-a-guided-question-flow-ab-test

## Hypothesis

If filtering is guided by questions such as who the product is for and what the budget is, visitors who are unsure which options to tick may reach suitable products more easily. Product page visit rate may go up as a result, while visitors who already know what they want may be slowed by the extra steps.

## When to run this test

Filters selected and removed right away, combinations that often end in zero results, or support requests asking which product is right all show visitors struggling to decide what to tick in the filter panel. If the guided flow helps, its users reach a product page with less trial and error, and those who filter directly are not slowed down. If the cause is confusing filter labels or missing product attributes, fix those first, because a question flow cannot repair faulty product data.

## Primary metric

**Product page visit rate**: Compare product-page visits per eligible assigned visitor while keeping the classic filter panel available in both arms.

## Other metrics

- **Guided-flow completion rate**: Report completion among visitors who start the guided flow; keep the primary analysis on all assigned visitors.
- **Classic filter usage**: Track use of the classic filters, which remain available in both arms.
- **Zero-result rate**: Compare eligible searches that end with no results.
- **Time to reach a product**: Measure time using the same start and arrival events; report visitors who never arrive.

## What to check before and during the test

- **Path**: Compare the classic filter panel with one predefined guided question flow that has the classic panel as a fallback.
- **Skip**: Keep direct access to filters available before the questions are completed.
- **Questions**: Fix the number, order and wording of questions before launch.
- **Mapping**: Map equivalent answers to the same core attributes and result rules.
- **Visitor history**: Review new and returning visitors separately.

## What to watch when reading the result

- A higher guided-flow completion rate does not pick the winner; report it only among visitors who start the flow, as a diagnostic. The deciding metric is product page visit rate across all eligible assigned visitors; when you report time to reach a product, list visitors who never get there separately.
- Review the answer combinations that end with no results; each of those screens needs a way to edit answers or switch to the classic filters. A dead end drags down the variant's result even if the idea of a guided flow works.
- Leave answers out of the next visit in this test; keeping them without asking the user is a separate change. Read new and returning visitors separately, and treat the result as valid only for this question set and these mapping rules.
- An unclear difference may mean the effect is diluted, because only some assigned visitors use the flow. Recalculate the visitors needed per variant by entering the product page visit rate for everyone assigned and that diluted lift into the [sample size calculator](https://alidemirbas.com.tr/en/calculators/sample-size-calculator). Dividing the result by each arm's daily traffic shows how much longer the test has to run.

## Frequently Asked Questions

### Can the guided questions be mandatory?

No. Preserve direct filtering before completion and access to the classic panel in both groups. Equivalent answers must use the same filtering rules. Provide editing and exit routes for empty combinations.

### Can guided-flow completion determine the winner?

No. Report completion among flow starters diagnostically. Use product-page reach rate as the primary outcome.

### Can answers be reused on the next visit?

Not in this test. Persistence is a separate change requiring a defined purpose, retention period and user controls. Review new and returning visitors separately.